Model Predictive Control of an Underground Mine Cooling Water Operation

Deep-level mines require energy-intensive cooling processes to maintain a workable environment up to 5 km underground, where temperatures reach 65 degrees C. The investigated cooling circuit consists of storage dams, cooling towers and five refrigeration plants, accounting for a large percentage of the plant's energy usage. The mine electricity supplier has implemented time-of-day tariffs for demand-side management. While high-level control is currently implemented on the cooling circuit by Mintek, the economic impact of time-of-day tariffs is not accounted for, partly due to the lack of accurate models and parameters describing the system. It is estimated that scheduling fridge plant operation based on the downstream underground needs and the variable price of electricity could save up to ZAR16 million (similar to USD 840 000) per year in energy costs according to a preliminary investigation done by Mintek. A model of the cooling circuit was developed from first principles using mass and energy balances. The model was calibrated using existing plant data and the parameter values regressed for and evaluated by means of an identifiability analysis using the profile likelihood method. The use of noisy plant data caused an integrator response and created the need for a Constrained Kalman Filter to bring the predictions back within the range of the measurements for regression. The validated plant model was used to implement a model predictive controller (MPC) in simulation on each unit, with an economic scheduling objective to be implemented in future work. Copyright (c) 2024 The Authors. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)